Evidence on Contagion in Management of Earnings

In this paper, we study peer firm behavior to ascertain (i) whether exposure to financial misrepresentation fuels similar behavior or deters it among peer firms; and (ii) what factors increase or decrease the likelihood of contagion? The prior years have witnessed a remarkable increase in news about corporate misconduct including fraudulent financial reporting at companies such as Enron, WorldCom, and Tyco, the collapse of Arthur Andersen on allegations of lax or corrupt audit work, tax shelters structured by KPMG to assist clients in minimizing tax obligations, and the revelation of the $50 billion Ponzi scheme run by Bernie Madoff. Given such frequent exposure to corporate misconduct, an important question relates to the impact of such encounters, if any, on managers of peer firms. Does such exposure deter misconduct among managers of peer firms? Or, does it encourage peer firms to also engage in questionable behavior?

Announcement of misconduct by other firms, and the consequences they face, is likely to enable peer firms to learn about (i) the details of the misconduct (for instance, the use of early revenue recognition or the nature of the restating firm); and (ii) the costs of engaging in questionable accounting practices. If the target restating firm, upon discovery of misrepresentation, faces little or no regulatory enforcement then a peer firm is likely to conclude that the costs of managing earnings are low. Lower expected cost of misconduct implies that managers of peer firms may rationally choose to adopt these practices as the benefits of such behavior outweigh the costs (Becker 1968). However, contagion is only likely to arise if the perceived costs borne by the misreporting firm are low. If the misreporting firm is subject to litigation or SEC enforcement, peer firms are likely to shun questionable practices, leading to the phenomenon we refer to as deterrence.

We investigate whether peer firms choose to engage in misrepresentation when target firms in their industry or geographical location publicly reveal misrepresentation through a restatement during the years 1997-2008. An average period of 2.6 years elapses between the beginning of misrepresentation and its public revelation through a restatement announcement in our sample. If the peer firm begins earnings management during this period, when there is no public knowledge of misconduct at the target firm, it is likely due to similar economic pressures to misrepresent or from private knowledge of such practices. Private information about earnings management could be obtained from a common lax auditor’s office (Francis and Michas 2013) or common board member (see Bizjak, Lemmon and Whitby [2009] and Chiu, Teoh and Tian [2013]). One would expect that the public acknowledgement of non-GAAP accounting practices in the restatement announcement would lead to cessation of misreporting among peer firms. Hence, we rely on a significant increase in the likelihood that a peer firm begins misrepresenting after the announcement of a restatement by the target firm as evidence consistent with the contagion hypothesis.

To focus on restatements that are more substantial likelier to be imitated, we eliminate restatements that involve an increase in net income. That is, we only include restatements that involve inflated net income during the violation period. To ensure access to a long time series of restatements, our sample comes from two sources: (i) 179 restatements that report decreased net income over the period 1997-1999 obtained from the General Accounting Office list (GAO); and (ii) 2,197 restatements for the period 2000-2008 that decrease net income collected from the Audit Analytics database. We control for contemporaneous adoption where earnings management by peers starts after the beginning date of the target firm’s violation period but before its public announcement of the restatement. Our measure of contagion captures only peer firms that begin managing earnings after the public announcement of the restatement by the target firm.

Given the significant changes in capital markets during our sample period, most notably the passage of the Sarbanes Oxley (SOX) act in 2002, we analyze three year sub-samples of our data. In the pre-SOX period of 1997-1999 (GAO sample) and the 2000-2002 (Audit Analytics sample) we find that a firm has a significantly higher probability of beginning earnings management if a higher fraction of its industry and its geographical neighborhood, measured as the metropolitan statistical area (MSA) the firm is located in, has already revealed they managed earnings via a public restatement announcement in the prior year. This evidence of contagion within industry and MSA explains a significant portion of variation in the peer firm’s decision to begin managing earnings.

However, such contagion, both at the industry and the MSA level, disappears in the three year period following the passage of SOX (2003-5) presumably because of the implementation of a stricter regulatory framework. Thus, there is evidence consistent with the deterrence hypothesis in the post-SOX period. Interestingly, there is some evidence that industry-level contagion reappears in the 2005-8 period. We conjecture that questionable reporting practices resurface after the initial sting associated with post SOX regulatory regime has abated.

We also study the channels that increase or decrease the likelihood of contagion. We begin by documenting that contagion among peer firms is more likely to be observed in the same account as the one restated by the target firm. Peer firms learn not only about the costs of misconduct but also about the specific accounting principle and its application within the industry. We test for potential contagion in similar accounting practices by examining restatements that involve revenue manipulation, expense accounts and asset, inventory or restructuring. We find that restatements that involve these specific accounts are associated with earnings management among the industry peers in the same accounts suggesting that similar accounting treatments potentially diffuse among peer firms.

Moreover, we exploit three sources of cross-sectional variation in the nature of the restatements to better understand the forces that impact contagion. First, we look at both the public enforcement through SEC investigations and private enforcement though class action litigations. The presence of public and private enforcement actions increases the cost of managing earnings for the restating firms and is likely to discourage peers and therefore should not be associated with contagion. Consistent with our conjecture, we find that restatement announcement accompanied by SEC enforcement actions or class action litigations are not associated with contagion in earnings management. Such deterrent effects are observed for both industry and MSA peers.

Second, we study the characteristics of the restatement. Extreme restatements involve substantial manipulation and are likely to be perceived as too severe for peer firms to imitate. Hence, we expect extreme restatements to be associated with no contagion. We classify restatements as severe if they fall in the most severe quartile of restated negative net income scaled by total assets. As expected, there is little evidence of contagion following severe restatements.

Lastly, we look at the characteristics of the restating or the target firm. In particular, large and visible firms are more likely to cause peer firms to adopt such practices or lead to contagion. Consistent with this conjecture, we find evidence of significant contagion when the target firm is above the median size in the industry. Along similar lines, we find that disclosure prominence also impacts contagion. Restatements made via a press release are more likely to be associated with contagion, among both industry and MSA peers.

One alternate explanation for the results is that they pick up some omitted industry factor that is correlated with payoffs to earnings management. Three sets of findings negate such a possibility. First, our results are robust to (i) explicit controls for industry overvaluation and industry structure; and (ii) the inclusion of contemporaneous adoption, or the adoption of similar practices prior to the public disclosure, that capture any residual industry level trends in earnings management practices. Second, evidence of significant contagion at the MSA level, not just at the industry level, also points against the results being entirely attributable to omitted industry factors. Third, contagion varies within an industry and an MSA depending on the presence or absence of enforcement and several predictable characteristics of the restating firm and the restatement itself. Such within industry or MSA variation in contagion is also not consistent with an omitted industry or MSA variable driving the results. Earnings management is likely to be pro cyclical and another possibility is that our results reflect economy wide or macro trends in the adoption of aggressive accounting practices. As our results are robust to the inclusion of year effects as well as stock returns for the firm and the industry’s book-to-market ratio, it is unlikely that our findings are entirely due to correlation of earnings management with business cycles.

The existence of contagion in earnings management after an initiator announces a restatement and the factors that seem to encourage (e.g., same accounts, if the initiator is a larger firm, prominent disclosure of the initiating firm’s announcement) and deter such contagion (e.g., SEC action or a lawsuit against the initiator) are new to the accounting and the financial economics literatures. In particular, these results could interest enforcement agencies such as the SEC. For example, evidence on contagion in aggressive corporate reporting potentially implies large differences across industries and geographical areas in the marginal productivity of enforcement spending by the SEC. We are also among the first to document that SEC enforcement and class action lawsuits deter earnings management. Finally, we are perhaps the first paper to raise the possibility that earnings management resumes once the sting associated with the vigorous SOX related enforcement wears off in the 2005-8 period.

The preceding post comes to us from Kevin Koh, Assistant Professor at the Nanyang Business School, Singapore, Simi Kedia, Professor at Rutgers University and Shiva Rajgopal, the Schaefer Chaired Professor of Accounting at the Goizueta Business School at Emory University.  The post is based on their recent paper, entitled “Evidence on Contagion in Earnings Management”, which is forthcoming in The Accounting Review and available here.